>>1038072
This is correct. You can double the number of transistors in a given area every 18 months or so. It's slipped closer to 2 years.
But this has never meant "double" the performance. There are many other design factors involved.
Voltage
Clock
Arrangement
That last one is big. Dedicating chunks of space to dedicated circuits for calculating specific things is a trade-off with the others
Every CPU is a series of trade-offs
But over time we've been able to design both faster circuits at the same clock, circuits with lower power draw, and circuits that can clock higher
Modern CPUs are combination of all these technologies
But they all rely on 2D lithography to build them
And transistors can only be built so small
Even if we could have a transistor be the size of a single atom, there's a limit there. And realistically you need "walls" that are many atoms thick.
We'd need to tame quantum mechanics (in at least a limited form) to go any smaller
Even then, how easily can we manufacture hardware like this?
Not very
Intel has a lot of intelligent people working in their labs and they are doing the best they can to continue performance gains
But I for one am looking forward to the return of large, bulky machines and concerns over performance
If every developer today put more effort into optimizing their code then we'd probably save billions on electric bills each year, cumulatively; so many cycles are wasted brute-forcing a problem with a languages that have massive overhead
Of course, maybe (((they))) will just insist we use phones as ultra-thin clients (literally) and offload all processing to server farms where we won't have to see ugly, bulky hardware.
And the masses will hand their phones in for a glorified Remote Desktop session so that Apple and Google can just stream their device to them